Results obtained from the plots of a field experiment may be subject to’ representational errors'; that is, they may not exactly reproduce the results that would be expected if the same treatments were applied to whole fields. In particular, the treatment applied to one particular plot often has the potential to affect the adjacent plots; such an effect may be positive or negative according to circumstances, and its magnitude cannot easily be estimated. This paper shows, on theoretical grounds and by simulations, that nearest-neighbour (NN) analysis cannot be relied upon to mitigate the effects of such inter-plot interference, even when the residual variance of the NN analysis is substantially, and by normal criteria significantly, smaller than that of conventional (randomized block) analysis. In certain circumstances, NN analysis of data affected by interference produces greater average bias in estimated treatment differences than conventional analysis.

At four pairs of sites in the UK, two cultivars of wheat, bred either for autumn (cv. Mercia) or for spring (cv. Tonic) sowing were sown in November 1988 and 1989 and their yield response to springapplied fertilizer N assessed over the range 0–320 kg/ha N. Grain samples were taken at harvest and analysed for those qualities regarded as indicators of suitability for milling. In five experiments, the grain yield of Mercia was up to 1 t/ha greater than that of Tonic, but overall there was no consistent difference between the two cultivars in the optimum economic yield (Yopt)or in the amount of fertilizer N needed to produce it (Nopt). Both cultivars produced c. 17·5 kg of grain per kg of fertilizer N applied up to Nopl. In the five experiments in which Tonic yielded less, it produced grain of greater N concentration than Mercia when given the same amount of fertilizer N. Differences in grain N concentration, while small (< 0·3%), were significant when analysed across sites. In six experiments both cultivars produced grain of 11 % protein at Nopt. There were no consistent or worthwhile increases in specific weight, Hagberg falling number (HFN) or response to the sodium dodecyl sulphate (SDS) sedimentation test of either cultivar from applying more fertilizer N than Nopl. The average specific weight of Mercia grain was 2·5 kg/hi greater than that of Tonic, while Tonic gave larger SDS values of c. 8. There was little difference in HFN. It was concluded that the yield requirement for fertilizer N was similar for these two cultivars of wheat when sown in late autumn, and there was no need to apply differential fertilizer N applications to the cultivars to produce grain of the quality needed for breadmaking. In four experiments the application of plant growth regulator (PGR) increased the yields of both cultivars by c. 0·5 t/ha.

At 14 sites in the UK, spring wheat (Triticum aestivum) cv. Tonic, was sown on three or four dates at each site between October and March in the 1988/89, 1989/90 and 1990/91 seasons. Responses to spring-applied fertilizer N over the range 0–320 kg/ha were determined. Earlier sowing did not increase uptake of soil N by the crop. Fertilizer N increased grain N offtake by between 25 and 140 kg/ha and yield by between 0·3 and 5·5 t/ha, although grain yield was less responsive to fertilizer N at later sowing dates. Apparent recovery of fertilizer N (AFR) also decreased as sowing was delayed but there was no effect of delayed sowing on the amount of grain produced from each kg of fertilizer N recovered. Because fertilizer N recovery decreased with later sowing, the amount of fertilizer N needed to produce the optimum economic grain yield was not reduced. Neither AFR nor optimum fertilizer (Nopt) was related to optimum yield. Regression of Nopt on the difference between optimum yield and yield without fertilizer N (△y) explained 77% of the variance in Nopt. There was an inverse relationship between △y and soil mineral N (SMN) in spring; regression of △y, on SMN in spring accounted for 29% of the variance in △y Current advisory systems which adjust economic fertilizer N recommendations according to anticipated yield are not justified by these results. Moreover the adjustments made, based on yield expectation, appear about three times as large as those needed to minimize residues of fertilizer N left unrecovered by the crop and to reduce the risk of nitrate leaching in the following winter.

Maize (Zea mays L.) hybrids with different productivity per unit approximate leaf projected area (ALPA) were compared in China, 1990–92. The results showed that the selection index (SI) values of all the hybrids tested decreased with an increase in population density. Nevertheless, the hybrids with high SI values at low plant densities also had relatively high SI values at high plant densities. Thus, high SI hybrids could be selected at any of these densities. Although no significant yield increase was obtained by using high SI hybrids at plant densities of 52500 plants/ha or lower; significant increases in yield were obtained for high SI hybrids at > 67 500 plants/ha. Leaf area was one of the parameters necessary for determining SI, and this could be predicted with a skewed bell-shaped function simply by measuring the length and breadth of the largest leaf and counting the total number of leaves per plant.

Four plant species were compared in each of three harvest periods (in August/September) in 1991 and 1992 at Aberystwyth: white clover (Trifolium repens L.), rape (Brassica napus L.), spurrey (Spergula arvensis L.) and perennial ryegrass (Lolium perenne L.). Plant physical structure was considered in relation to rate of intake by sheep, chewing activity and the effectiveness of chewing in breaking the diet into particles.

White clover had a much lower proportion of cell wall than perennial ryegrass, but the rate of intake and the number of chews per min were similar for the two species. White clover petioles broke down into long, thin particles, similar in size and shape to those derived from perennial ryegrass leaf sheaths; many of the clover petioles were not split longitudinally by chewing, in contrast to the ryegrass sheaths. A white clover leaflet was typically broken into about 20 blocky particles, whereas a petiole of similar weight was broken into only about three particles. Veins were close together in ryegrass leaf sheaths and blades, particularly the latter; approximately one in seven strips of weaker tissue between veins was ruptured by chewing leaf sheaths and one in 16 in the case of leaf blades, in each case resulting in particles of c. 2 mm width. Rape had a low proportion of cell wall and a low proportion of vascular tissue in its leaf blades, petioles and stems. Rape leaf blades were eaten quickly, but the stems were eaten slowly. The length and width of particles derived from rape leaf blades were very similar to those of particles derived from white clover leaflets. Spurrey had a high proportion of cell wall and was low in in vitro digestibility, but the rates of intake and chewing were high and relatively few chews were required per g of dry matter ingested. The vascular bundles in the spurrey stems were only half the thickness of the bundles in white clover petioles; pieces of spurrey stem were typically broken at about two places along their length and were not split during eating.

The study illustrates the wide variation in plant anatomy among species which can be available to herbivores and some effects of the abundance, thickness and orientation of vascular bundles on rate of intake, chewing activity and the size and shape of particles produced by chewing.

Potato seed tubers infected or not infected with gangrene (Phoma foveata) were planted at Rothatnsted in 1987 to measure the effect of the disease and of neighbouring plants on yield. The experimental design was constructed so that the effect on growth of six adjacent plants (two nearest neighbours in each direction within rows and one nearest neighbour in each direction across rows) could be estimated for each plant. Total yield, ware (> 150 g) yield and tuber number from individual plants were affected most by the disease but also, in decreasing importance, by the two plants on either side within the same row (first neighbours), the two plants adjacent to the first neighbours (second neighbours) and the two adjacent plants in the rows on either side. Yield and tuber numbers increased as the different combinations of neighbouring plants contained increasing proportions of plants from diseased seed and missing plants; plants compensated for decreasing competition. Tuber size distributions showed that numbers of ware tubers decreased with increasing competition whereas numbers of small tubers were less affected. The fitted model was used to predict yields from crops planted with different proportions of diseased or missing seed tubers.

In 1990–92, field experiments were performed at Broom's Barn Experimental Station to study the effect of 5 years' repeated straw incorporation on sugarbeet. Straw incorporation had no effect on plant population density. Processing quality was reduced by incorporated straw but N had a much larger effect. The effect of incorporated straw on the mineral N content of the soils and N uptake by beet was inconsistent, and this may be related to the amount of soil mineral N present when the straw was incorporated. The efficiency of fertilizer use was unaffected by straw incorporation. On Broom's Barn soils when straw was incorporated, the optimal economic N dressing was c. 120 kg N/ha, and in unincorporated plots it was c. 100 kg N/ha. At the optimal economic N rate, incorporated straw increased beet yields.

Data from the F1, F2 and F3 generations derived from four F1 hybrids of sunflower were analysed to determine the genetical control of eleven agronomically important traits namely, flowering time (FT), final height (FH), seed set (SS), fresh seed weight (FSW), dry seed weight (DSW), 25-seed weight (SSW), head diameter (HD), head angle (HA), neck width (NW), Botrylis infection (BI) and percentage oil content (PO), with the objective of assessing their potential for producing superior recombinant lines. Comparisons of means revealed significant differences between the F: generations of crosses for all the characters except FT. Model fitting indicated that these differences were attributed either to variation in the magnitude of the dominance component [h], or the projected mean of the Finfin; generation, m. Further, all the variation in the generation means was accounted for by the m and [h] parameters and epistasis was shown to be non-existent for almost all the traits. Analysis of the second degree statistics revealed the presence of additive genetic effects and most of the characters showed moderate heritability except for HD, which was highly heritable. All the crosses were predicted to produce significant numbers of transgressive segregants and consequently were expected to yield superior inbred lines and second cycle hybrids either to be released as varieties as such or to be utilized in future breeding work. The overall differences between the crosses, however, were not highly pronounced and therefore it was not possible to identify a cross which would yield markedly superior recombinant lines and F1 crosses compared to the others.

A 2–year study (1990 and 1991) was conducted at the ICRISAT (International Crops Research Institute for the Semi-Arid Tropics) Sahelian Centre, near Niamey, Niger, to select groundnut cultivars tolerant to drought and to examine selection techniques. Thirty-six cultivars known to vary in yield potential were grown under rainfed and irrigated conditions. Crop growth rate (C) and partitioning co-efficient (p) were estimated from phenological and final harvest data. The correlation between years was greater for partitioning than for pod yield (implying a higher heritability for p than for yield). Tolerance as determined by a drought susceptibility index for pod yield (SY), crop growth rate (Sc) and partitioning (Sp) to reproductive sinks showed thirteen cultivars as drought tolerant for either C or p or for both. The Sahelian cultivars 796, 55–437 and TS 32–1 were the most consistent for drought tolerance. Partitioning was the most important yield component affecting yield variation among cultivars.

A factorial experiment on Vicia faba L. cv. Alameda with three plant densities at three levels of nitrogen fertilization was done for three years (1984–87) at Granada. Yield varied markedly from 403 to 579 g/m2 between seasons. The mean increase in plant population density from 10 to 16 plants/m2 raised seed yield, but a further increase to 21 plants/m2 did not increase yield significantly. There was, however, considerable variation in the response to increasing density (Dl to D3) within each N level (N0, N1 and N2). Application of 30 kg N/ha at the start of flowering increased mean seed yield only at the lowest plant density and in 1986, but 60 kg N/ha increased yield by 135% on average over the 3 years. The highest N application rate resulted in consistently higher overall yields at each plant density. Leaf area index (LAI) showed a significant increase at 20 seeds/m2 compared with that at 10 seeds/m2 but there was no further increase at 30 seeds/m2. Harvest index (HI) was mainly affected by the annual weather conditions, and ranged from 47–4 in 1986 to 62–6 in 1987. HI was related to the mean temperature at flowering (April).

In a field experiment with transplanted rice (Oryza sativa L.) during the wet season of 1990 at the Central Soil Salinity Research Institute, Karnal, India, the effects of four rates of N application (0, 80, 120 and 160 kg N/ha), two methods of urea application (before puddling, M,, and after puddling, M2) and two presubmergence periods (none, S0, and 1 week, S1,) were evaluated on ammonia volatilization losses in a sodic soil (pH 9·0, exchangeable sodium 29%). Cumulative ammonia volatilization losses were 1·43, 18·00, 22·54 and 28·29 kg/ha following applications of 0, 80, 120 and 160 kg N/a respectively; 17·08 and 18·01 kg/ha for treatments M1, and M2 respectively; and 18·63 and 16·34 kg/ha for treatments S0 and S1 respectively. Flood-water NH4+ and pH increased during days 1–4 and then there was a sharp decline on days 4–9, followed by a gradual decline. The highest flood-water NH4+ and pH values were found for the treatments which received urea at the rate of 160 kg N/ha, after puddling (M2) and with no presubmergence period (S0).

Grazing sambar and red deer in New Zealand were offered a free choice of seven different plant species (forage legumes, browse and grasses) in 1992 and 1993 and dietary preference rankings were determined. Nutritive value of plants on offer and diet selected, plant height, plant species purity and stem diameter selected (browse only) were also determined. Total nitrogen (N) and organic matter digestibility (OMD) were highest for red clover, lowest for grasses and intermediate for browse species (willow, poplar and lupin). Top dietary preference ranking was willow for sambar and red clover for red deer in both years, with Yorkshire fog and prairie grass being lowly preference ranked with both deer species. Sambar selected willow stems up to 38 mm in diameter and poplar stems up to 54 mm in diameter. When the plants were grouped into browse, grass and forage legume categories, both deer species showed a similar preference ranking for grasses. Relative to grasses, sambar showed a strong preference for browse and a low preference for forage legumes, whilst red deer showed a strong preference for forage legumes of high nutritive value and a very low preference for browse. It was calculated that sambar selected a total diet higher in condensed tannins and lignin but lower in nitrogen than that selected by red deer, with similar values for total fibre and OMD. Differences in dietary preference between the two deer species may be linked with the greater ability of sambar deer to neutralize some plant secondary compounds and their more efficient rumination pattern compared with red deer. Both sambar and red deer can be classified as intermediate feeders, having a similar preference for grasses, but differing preferences for forage legumes and browse.

During 1991/93, young sambar (5 months old) and red deer (7 months old) were confined indoors in individual pens in New Zealand and fed a pelleted concentrate diet (12 MJ ME/kgDM; 2·9% N) ad libitum for 21 months. Seasonal patterns of voluntary feed intake (VF1), liveweight gain (LWG), scrotal circumference and plasma concentrations of prolactin (PRL), luteinizing hormone (LH), testosterone (T) and progesterone (P) were compared using five stags and three hinds of each species.

Red deer showed a strong pattern of seasonality, with high VFI and LWG in summer and low VF1 and LWG in winter and with peak plasma T and scrotal circumference in stags in early autumn. Compared with red deer, sambar showed weaker seasonal patterns of VFI and LWG, with maximum values in autumn and minimumvalues in spring. Over a complete 12-month cycle, sambar deer gained similar amounts of liveweight to red deer but consumed substantially less feed, thus demonstrating a more efficient conversion of feed tobodyweight. Metabolizable energy (ME) requirements for both maintenance and gain were substantially lower for sambar than for red deer. Scrotal circumference and plasma T values in sambar stags attained their highest values during late autumn, winter and spring, but with a lower magnitude than peak values for red stags. Plasma PRL concentrations were seasonal in both species, with highest values in summer and lowest values in winter. Rapid increase of plasma P was first detected in red hinds in autumn and sambar hinds in spring when they weighed 96 and 90 kg respectively, and were aged 17 and 14 months. Rapid increase of plasma T was first detected in red stags in early autumn and sambar stags in mid-autumn whenthey weighed 117 and 101 kg, and were aged 16 and 15 months respectively.

It was concluded that sambar deer had endogenous cycles of VFI, body growth and hormone secretion, which were of lesser amplitude and with different seasonality from those of red deer. Young sambar deer were more efficient feed converters than red deer, and attained sexual maturity at an earlier age and lower liveweight.

Isoenergetic diets containing grass silage and concentrates in the ratios 75:25; 50:50 and 30:70 on a dry matter (DM) basis were offered to 42 beef heifers, which were initially 361 kg liveweight, for 115 days in 1987 at Hillsborough, Northern Ireland. Half of the animals given each forage concentrate ratio were given one of two crude protein (CP) intakes in a 3 × 2 factorial design. The silage contained 152 g CP/kg DM, 94 g ammonia-N/kg total N and 734 g digestible organic matter/kg DM. Concentrates were based on barley and soyabean meal and were formulated to equalize CP intakes for the three forage: concentrate ratios. For diets containing 75, 50 and 30% silage, metabolizable energy intakes were 94, 92 and 87 MJ/day; liveweight gains 0·99, 1·07 and 1·12 (S.E. 0·051) kg/day; carcass gains 0·61, 0·64 and 0·67 (S.E. 0·033) kg/day; carcass lean concentrations (estimated from dissection of fore-rib joints) 636, 642 and 648 (S.E. 4·7) g/kg and carcass fat concentrations 204, 199 and 194 (S.E. 5·3) g/kg. Increasing protein intake did not affect performance, but increased carcass fat concentration at all three forage concentrate ratios. It was concluded that increasing the proportion of concentrates in grass silage-based diets increased carcass weight gain per MJ of metabolizable energy consumed and tended to reduce carcass fatness. Contrary to previous findings with young growing cattle, increasing protein intake did not alleviate the problem of greater fat deposition in finishing cattle given high-silage diets, but rather significantly increased carcass fat concentration.

In ruminants, the urinary excretion of purine derivatives (PD) reflects the absorption of microbial purines and can be used as an index of microbial protein supply. The objective of this study, carried out in Aberdeen, 1992, was to examine whether PD concentrations in spot urine or plasma samples vary diurnally during a given feeding regime and if they reflect differences in daily PD excretion induced by varying feed intake. Sixteen sheep were offered ad libitum one of four diets (fresh weight basis, the rest of each diet being minerals and vitamins): (1) 99·9% lucerne (pelleted); (2) 50% hay, 30% barley, 9% fishmeal and 10% molasses; (3) 72% straw, 7% molasses and 20% molassed sugarbeet pulp; and (4) 97% barley. Measurements were made for 1 week after a 2-week adaptation period. Urine was collected daily on days 1–4 and hourly on days 5–7. Hourly urine collection was achieved using a fraction collector. Plasma samples were collected hourly from 09·00 to 17·00 h on day 4. Feed intake varied considerably (347–1718 g DM/day) between diets and between animals. Daily excretion of PD (7·1–22·6 mmol/day) was linearly related to DM intake (r = 0·85, n = 16), and so was the microbial N supply (3·9–19·5 g N/day) estimated from daily PD excretion (r = 0·87). In hourly urine samples, the ratio of PD:creatinine concentrations showed no significant difference between sampling times, and was linearlycorrelated with the daily PD excretion (r = 0·92). Similarly, plasma PD concentration also showed little diurnal fluctuation. Glomerular filtration rate (GFR) increased with feed intake. Plasma PD was not well correlated with daily PD excretion in urine (r = 0·57). The tubular load of PD (plasma PD × GFR) was better correlated with the daily excretion (r = 0·80). It appears that when sheep are fed ad libitum, PD in spot urine may provide a practical indicator of microbial protein supply status.

Studies were conducted in 1989 at the University of Zaragoza, Spain, to determine the value of undegradable dry matter and neutral detergent fibre (uDM and uNDF) forestimating apparent dry matter digestibility (DMD), with sheep diets based on untreated or ammonia-treated barley straw (US or TS) offered alone or supplemented with 300 or 600 g/day of grass hay (H), barley (B) or sugarbeet pulp (P). Marker recovery in faeces was 107·9±1·90 and 96·4±1.43% for uDM and uNDF, respectively. Faecal uDM excretion was overestimated, especially with TS as sole feed and in B-supplemented diets (116%), whereas underestimation of uNDF excretion was observed mainly in unsupplemented US (92%), supplemented TS (92%) and in diets including H (91 %). Digestibility estimations from uDM or uNDF concentration in feeds and faeces (DuDM or DuNDF) were affected in the same way as DMD by the dietary sources of variation. Correlation coefficients between in vivo digestibility and its estimates with both internal markers were 0·80, and differences between DMD and DuDM and DuNDF were –2·5±0·14 and 2·2±0·13 percentage units, respectively. The suitability of both markers for predicting apparent digestibility, and the choice between them, would depend on diet characteristics.

The aim of this experiment was to determine whether the thermal responses ofsheep exposed to a hot environment were similar in different years. Rectal temperaturesofsheep were measured during the hot months (between November and March) of three consecutive years (1990/91, 1991/92, 1992/93) in a flock of Merino sheep (n = 151) in the semi–arid tropics of northern Australia. Mean daily maximum temperatures in these months were 36±0·4, 39±0·3 and 37±0·3°C in Years 1, 2 and 3, respectively. In Year 1, rectal temperatures were measured on 3 consecutive days;the repeatability was 0·47 at 08·00 h and 0·64 at 16·00 h. In Year 1, two sub-groups were identified according to measurements of rectal temperature, low body temperature(LBT; n = 35) sheep, whose rectal temperature was ≤ 39·8°C at 16·00 h on each of the 3 consecutive measurement days, and high body temperature (HBT; n = 32) sheep, whose rectal temperature was ≥ 39·9°C. The rectal temperatures of LBT and HBT sheep were then measured during the subsequent 2 years. Rectal temperatures of HBT sheep were significantly higher than those of LBT sheep at both 08·00 and 16·00 h, in both Year 2 and Year 3. Rectal temperatures at 16·00 h in Year 1 were correlated (P < 0.001) with rectal temperatures at 16·00 h in both Year 2 (r = 0·79) and Year 3 (r = 0.82), respectively. It was concluded that measurement of rectal temperature in sheep exposed to a hot environment in any oneyear is an accurate index of their rectal temperatures during subsequent years.

This group, which is concerned with the applications of mathematics to agricultural science, is sponsored by the Biotechnology and Biological Sciences Research Council. It was formed in 1970, and has since met at approximately yearly intervals in London for one-day meetings. The twenty-sixth meeting of the group, chaired by Professor R. J. Wilkins of the Institute of Grassland and Environmental Research, North Wyke, was held in the Wellcome Meeting Room at the Royal Society, 6 Carlton House Terrace, London on Friday, 7 April 1995, when the following papers were read.